Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
A cohort of patients with intraductal growth-type intrahepatic cholangiocarcinoma (IG-ICC) and its precursor lesions, collectively termed intraductal papillary neoplasm of the liver (IPNL), was characterized with respect to demographics, clinical manifestations, perioperative management, long-term survival, and molecular features associated with carcinogenesis. A total of 122 patients with IPNL types 1 through 4, 108 patients with non–IG-ICC and 210 patients with hepatolithiasis alone were studied. Expression of CDX2, TFF1, MUC1, MUC2, MUC5AC, EGFR, and p53 was determined by using immunohistochemistry. Females predominated in those with hepatolithiasis alone and IPNL. The mean age of patients with hepatolithiasis alone was 6 to 8 years younger than that of those with IPNL. The association with hepatolithiasis in patients with IPNL types 1 and 2, IPNL types 3 and 4, and non–IG-ICC was 100%, 79%, and 64%, respectively. Mucobilia, anemia, and elevated serum carcinoembryonic antigen levels were helpful in distinguishing IG-ICC and its precursor lesions. The mean survival of patients with IPNL type 3, IPNL type 4, and non–IG-ICC was 55.5 months, 36.9 months, and 15.8 months, respectively. The incidence of expression of CDX2 and TFF1 was maximal in IPNL type 3. Expression and cellular distribution of MUC2 and CDX2 were similar. MUC5AC was strongly expressed in all patients with IPNL; EGFR and p53 were rarely expressed in patients with IPNL. In conclusion, hepatolithiasis appears to be a precipitating factor in the development of IPNL. Signs of mucobilia were specific for the diagnosis of IPNL. Expression of CDX2 and MUC2 are helpful in differentiating IPNL and non–IG-ICC. Significant differences in survival associated with the various lesions studied warrants a more aggressive surgical strategy in their management. (HEPATOLOGY 2005;42:657–664.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Intrahepatic cholangiocarcinoma (ICC) is a primary adenocarcinoma originating from the second or higher order branches of hepatic bile ducts. The incidence of ICC exhibits wide geographical variations that range from 5% to 30% of primary liver cancers.1 Various predisposing factors, including liver fluke infestation, hepatolithiasis, primary sclerosing cholangitis, choledochocyst, and Caroli's disease, have been implicated.1–3 In Taiwan, ICC was found in 5% to 10% of patients with hepatolithiasis.3, 4 Based on gross morphology, ICC can be divided into three types (mass-forming, periductal-infiltrating, and intraductal growth [IG]-type).1 The IG-ICC is the least common type but is associated with the most favorable outcome.1, 5 Biliary papillomatosis, intraductal growth-type cholangiocarcinoma (IG-ICC), and mucin-producing ICC constitute essentially the same disease entity.5–7 This category of papillary biliary neoplasm resembles intraductal papillary mucinous neoplasm of the pancreas histologically.8 We recently proposed the term intraductal papillary neoplasm of the liver (IPNL) for such lesions, which include both IG-ICC and its precursor lesions.6, 9 Phenotypes of gastrointestinal metaplasia, such as colon-like metaplasia, goblet cell metaplasia, paneth cell metaplasia, and foveolar metaplasia, are typically found in IPNL but not in non–IG-ICC (mass-forming and periductal infiltrating types); these findings imply that these two types of lesion are associated with divergent cholangiocarcinogenesis pathways.
In this study, a cohort of patients with IPNL was characterized systematically with respect to demographics, clinical manifestations, perioperative management, long-term survival, and molecular features of carcinogenesis. Our findings should facilitate elucidation of the pathogenesis of IG-ICC, promote its diagnosis at an early stage, and improve the prognosis of this lesion.
IG-ICC, intraductal growth-type intrahepatic cholangiocarcinoma; IPNL, intraductal papillary neoplasm of the liver; TFF, trefoil factor family.
Patients and Methods
Definition and Pathological Descriptions.
IPNL is characterized by many frond-like papillary infoldings, which consist of columnar epithelial cells surrounding slender fibrovascular stalks that are supported by connective tissue from the lamina propria. Type 1 is an IPNL lined by biliary epithelium of low-grade dysplasia. Type 2 is an IPNL lined by biliary epithelium of high-grade dysplasia. Type 3 is an IPNL lined by in situ and microinvasive adenocarcinoma. Type 4 is characterized by types 2 and 3 biliary lesions and adenocarcinoma of variable invasiveness. Accordingly, IPNL types 3 and 4 correspond to IG-ICC associated with variable degrees of stromal invasion.
From 1977 to 2003, 85 patients with IG-ICC and 364 patients with non–IG-ICC underwent surgery in the Department of Surgery, Chang Gung Memorial Hospital. Of these, 72 (84.7%) of 85 patients with IG-ICC and 108 (29.7%) of 364 patients with non–IG-ICC underwent partial hepatectomy with curative intent. Of 72 patients with IG-ICC who underwent partial hepatectomy, 13 had IPNL type 3 and the remaining 59 had IPNL type 4. During the 5 years 1998 to 2003, 260 patients with hepatolithiasis were treated surgically in our institute; 50 (19%) of them were diagnosed as having IPNL, type 1 (n = 33) and type 2 (n = 17), after partial hepatectomy. The number of cases with hepatolithiasis with or without concomitant cholangiocarcinoma, who underwent partial hepatectomy from 1997 to 2003, was 378; the number with IPNL was 107. Accordingly, IPNL developed in 28.3% (107 of 378) of patients with hepatolithiasis. A total of 122 patients with IPNL types 1 through 4, 108 patients with non–IG-ICC, and 210 patients with hepatolithiasis alone constituted the patient population studied. The demographics, association with hepatolithiasis, symptomatology, serum tumor markers, radiological and gross anatomical features, and long-term survival were determined from examination of the patients' medical charts.
To investigate the expression of molecular markers in this cohort of patients, paraffin-embedded surgical specimens of all cases of IPNL (n = 122) and arbitrarily selected cases of non–IG-ICC (n = 30) and cases of hepatolithiasis alone (n = 30) were retrieved from the archives of our department of pathology. Of the 30 cases of non–IG-ICC, 12 had well differentiated adenocarcinoma, 9 had moderately differentiated adenocarcinoma, and 9 had poorly differentiated adenocarcinoma. Furthermore, those with hepatolithiasis alone (n = 10), IPNL type 3 (n = 6), IPNL type 4 (n = 8), and non–IG-ICC (n = 10) were arbitrarily selected for detection of CDX2 by Western blotting.
Immunohistochemistry of CDX2, TFF1, MUC 1, MUC2, MUC5AC, EGFR, and p53.
Four-μm-thick sections of formalin-fixed, paraffin-embedded tissue specimens were obtained. Immunochemistry of avidin-biotin-peroxidase complex was undertaken. Slides were deparaffinized in xylene and rehydrated in graded alcohols. Endogenous peroxidase was blocked using 0.3% hydrogen peroxidase in methanol. Slides were rehydrated in phosphate-buffered saline, pH 7.4. Various primary antibodies (CDX2, 1:100, monoclonal, Novocastra, Newcastle upon Tyne, UK; TFF1, monoclonal, 1:100, DAKO, Glostrup, Denmark; MUC1, 1:50, monoclonal, Novocastra; MUC2, 1:100, monoclonal, Novocastra; MUC5AC, 1:50, monoclonal, Novocastra; EGFR, 1:200, monoclonal, DAKO; and p53, 1:50 monoclonal, DAKO) were applied. Slides were washed in phosphate-buffered saline containing 1% Tween 20 for 10 minutes. Biotinylated goat anti-rabbit immunoglobulin (Vector Labs, Burlingame, CA) at a dilution of 1/500 was then applied. Tissue sections were reacted with 0.06% diaminobenzidine (Sigma Chemical Co., St. Louis, MO), counter-stained with hematoxylin, and dehydrated using graded alcohols; coverslips were applied with permount.
Tissue obtained from intraductal papillary mucinous tumor of the pancreas served as positive controls for CDX2, MUC2, MUC5AC, and TFF1 staining. No positive staining was obtained when primary monoclonal antibody was omitted or was replaced by normal mouse serum in negative controls for the staining procedures.
Double Staining for CDX2 and MUC2.
Serial sections of tissue samples embedded in paraffin, 4 to 5 μm thick, were mounted on gelatin-coated slides, dried overnight at 37°C, and deparaffinized using xylene. The detailed procedures of the double staining technique have been described elsewhere.10, 11 Tissue from an intraductal papillary mucinous tumor of the pancreas served as positive controls. No positive staining occurred when primary monoclonal antibody was omitted or was replaced by normal mouse serum in negative controls for the staining procedures.
Western Blotting for CDX2.
Frozen tissue was cut and homogenized in 200 μL ice-cold lysis buffer (10 mmol/L Tris-HCl, pH 7.5, 1 mmol/L MgCl2, 1 mmol/L EGTA, 0.1 mmol/L phenylmethylsulfonyl fluoride, 5 mmol/L β-mercaptoethanol, 0.5% Chaps, 10% glycerol). Total protein was measured using a Bio-Rad Bradford kit (Bio-Rad Laboratories, Hercules, CA); 30 μg total protein was subjected to 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis, and then transferred to a nitrocellulose membrane and blocked using nonfat dry milk in TBS-T (20 mmol/L Tris-HCl, pH 7.6, 150 mmol/L NaCl, 0.05% Tween 20) overnight at 4°C. The membrane was then probed with a primary antibody to CDX2, and incubated with a horseradish peroxidase–conjugated secondary antibody (goat anti-mouse horseradish peroxidase, Transduction Laboratories, Lexington, KY). Finally, the membrane was developed using an enhanced chemiluminescence system (Pierce, Rockford, IL). Quantification of the autoradiographs was undertaken using a Bio-Rad/GS 700 imaging densitometer.
Semi-quantification of CDX2, TFF1, MUC1, MUC2, MUC5AC, EGFR, and p53 expression detected by immunohistochemistry was scored according to the percentage of cells that were positive in an individual lesion: − (negative), 0%; 1+ (focal), 1% to 10%; 2+ (moderate), 11% to 55%; and 3+ (extensive), >56%. For simplicity, those lesions with a score of − or 1+ were considered to be immunonegative, and those with a score of 2+ or 3+ were considered to be immunopositive.
All continuous data were presented as means ± SD. Comparison of continuous variables among multiple groups was undertaken using one-way ANOVA, followed by Scheffe's method for individual paired analyses. Comparison of categorical variables among multiple groups was undertaken using the Pearson chi-square method. Kaplan-Meir curves and the log-rank tests were applied to compare survival between groups. Statistical significance was set at P less than .05.
The age and sex of patients with hepatolithiasis alone, IPNL, and non–IG-ICC are shown in Table 1. Females were more prevalent in those with hepatolithiasis alone and IPNL, whereas the prevalence of the two sexes in those with non–IG-ICC was similar. The mean age of patients with hepatolithiasis alone was 6 to 8 years younger than that of patients with IPNL and non–IG-ICC; the ages of patients with IPNL types 1 through 4 were similar. Thirty-nine (32%) of patients with IPNL and 29 (27%) of patients with non–IG-ICC had biliary surgery 1 or more times previously, usually for hepatolithiasis. Three patients had T-tubes for surveillance for recurrent hepatolithiasis; mucobilia was found during regular percutaneous fibrocholedochoscopic examination of the patient who had IPNL type 3 (n = 1) and the patients who had type 4 (n = 2).12
Table 1. Demographics of Patients With Cholangiocarcinoma and Its Precursors
Non-IG-ICC (n = 108)
Type 1 (n = 33)
Type 2 (n = 17)
Type 3 (n = 13)
Type 4 (n = 59)
Abbreviations: IPNL, intraductal papillary neoplasm of the liver; Non-IG-ICC, non-intraductal growth type cholangiocarcinoma. Values are expressed as n (%).
P < .01 versus that of hepatolithiasis alone group.
Symptomatology, the association of hepatolithiasis, and the presence of hepatitis markers, mucobilia, and serum tumor markers in patients with hepatolithiasis alone, IPNL, and non–IG-ICC are shown in Table 2. Right hypochondriac pain, fever, chills, and jaundice occurred more frequently in patients with hepatolithiasis alone and patients with IPNL than in those with non–IG-ICC. The association with hepatolithiasis in patients with IPNL types 1 and 2, IPNL types 3 and 4, and non–IG-ICC was 100%, 79%, and 64%, respectively (P > .05). Parasites are rarely associated with liver diseases in Taiwan. Anemia was more common in patients with IPNL types 3 and 4 than in those with IPNL types 1 and 2 (19% vs. 4%, P < .05). Mucobilia, which was identified by cholangiography, was more commonly found in patients with IPNL types 3 and 4 than in those with IPNL types 1 and 2 (38% vs. 18%, P < .01). Of the patients tested, elevated serum carcinoembryonic antigen levels were detected in none of 33 patients with IPNL types 1 and 2, and in 21% of patients with IPNL types 3 and 4 (P < .01). Of the patients tested, elevated serum CA19.9 levels were detected in 35% of patients with IPNL types 1 and 2 and 61% of patients with IPNL types 3 and 4 (P > .05).
Table 2. Symptomatology and Serum Tumor Markers in Patients With Cholangiocarcinoma and Its Precursor Lesions
IPNL Types 1 + 2
IPNL Types 3 + 4
(n = 210)
(n = 50)
(n = 72)
(n = 108)
NOTE. Values are expressed as n (%).
Abbreviations: IPNL, intraductal papillary neoplasm of the liver; Non-IG-ICC, non-intraductal growth0type cholangiocarcinoma.
P < .01 versus that of non–IG-ICC.
P < .05 versus hepatolithiasis and IPNL types 1 + 2, respectively.
Work-up for IPNL included liver ultrasonography, computed tomography, endoscopic retrograde cholangiography or percutaneous transhepatic cholangiography, and magnetic resonance cholangiography,13 in various combinations. During examination by endoscopic retrograde cholangiography or PTC, lumps of mucin were depicited as multiple, elongated linear, ovoid, or amorphous filling defects in dilated bile ducts, and constituted the most characteristic sign of IPNL.7 Sonography, computed tomography, and magnetic resonance cholangiography did not detect mucobilia, because the signal intensity of mucin is the same as that of water. IPNL types 1 and 2 could not be diagnosed preoperatively. Usually, IPNL types 3 and 4, with or without hepatolithiasis and mucobilia, could be identified preoperatively. However, generalized dilation of bile ducts, which was greater in hepatic segmental bile ducts, without any visable tumor, was found in three patients with IPNL type 3. Those with sustained jaundice or biliary tract infection were treated with antibiotics or biliary decompression, using either endoscopic nasal biliary drainage or percutaneous transhepatic biliary drainage. Operative methods used for patients with IPNL and non–IG-ICC are summarized in Table 3. Left hepatic lobectomy was commonly undertaken in patients with IPNL and non–IG-ICC. Common bile duct exploration was more frequently undertaken in patients with IPNL than in those with non–IG-ICC (98% vs 71%, P < .01). Surgical specimens from 3 patients with IPNL types 3 and 4 with IPNL type 4 had tumor-positive margins. One patient with IPNL type 3 and 3 with IPNL type 4 had dysplastic changes at the margins of surgical specimens. Hospital mortality in patients with IPNL types 1, 2, 3, and 4 and non–IG-ICC was 0%, 5.8%, 7.6%, 6.7%, and 4.8%, respectively (P > .05). All patients were followed up at 2- to 3-month intervals.
Table 3. Operative Methods in Patients With IPNL and Non–IG-ICC
Non-IG-ICC (n = 108)
Type 1 (n = 33)
Type 2 (n = 17)
Type 3 (n = 13)
Type 4 (n = 59)
Abbreviations: IPNL, intraductal papillary neoplasm of the liver; non–IG-ICC, non-intraductal growth-type cholangiocarcinoma; CBD, common bile duct.
Long-term survival of patients with IPNL and non–IG-ICC, which excludes hospital mortality, was analyzed (Fig. 1). At the time of this analysis, no liver disease–related death had occurred in those with IPNL types 1 and 2. One-, three-, and five-year survival rates were 100%, 33%, and 17% in patients with IPNL type 3; 56%, 32%, and 14% in those with IPNL type 4; and 45%, 9%, and 3% in those with non–IG-ICC, respectively. The mean survival times of patients with IPNL types 3 and 4, and non–IG-ICC were 55.5 ± 17.1 months (95% CI [22, 89]), 36.9 ± 6.3 months (95% CI [24.7, 49.2]), and 15.8 ± 1.8 months (95% CI [12.4, 19.3]), respectively. Although the survival of patients with IPNL types 3 and 4 was greater than that of patients with non–IG-ICC (log-rank test, P = .0009 and P = .0007, respectively), the survival of patients with IPNL types 3 and 4 was similar (log-rank test, P = .12). All 3 patients with IPNL type 3 and all 4 patients with IPNL type 4, who had surgical specimens with tumor-positive margins, died within 2 years without further resection of hepatic tissue. One patient with IPNL type 3 and 3 with IPNL type 4 underwent re-operation because of suspected locoregional recurrence of tumor an average of 11.3 months (range, 6-17 months) after the original partial hepatectomy. The patient with IPNL type 3 has survived for 3 years after further surgery without recurrence of disease. The other 3 patients had widespread dissemination of tumor at the time of further surgery. So far, there have been 9 deaths of patients with IPNL type 3 and 46 deaths of patients with IPNL type 4. Fifty-one of these 55 deaths were attributable to disseminated tumor.
Expression of CDX2, TFF1, MUC1, MUC2, MUC5AC, EGFR, and p53 in patients with hepatolithiasis alone, IPNL, and non–IG-ICC is summarized in Table 4. The immunohistochemical expression of CDX2 was detected in the nuclei of neoplasic cells (Fig. 2). The incidence of CDX2 expression was maximal in patients with IPNL type 3 (46%); CDX2 expression was detected in none of 30 patients with non–IG-ICC. Representative features of CDX2 expression in hepatolitiasis alone, IPNL, and non–IG-ICC, detected by Western blotting are shown in Fig. 3. CDX2 expression, detected by Western blotting, in hepatolithiasis alone, IPNL, and non–IG-ICC was similar to that detected by imunohistochemistry. Immunohistochemical expression of TFF1 occurred in the supranuclear cytoplasm as fine vesicles and in the apical region of biliary epithelial cells (Fig. 4). The incidence of TFF1 expression was maximal in IPNL type 3 (77%); TFF1 expression was detected in only 5 (17%) of 30 patients with non–IG-ICC. MUC1 was rarely expressed in patients with hepatolithiasis alone and IPNL; MUC 1 was strongly expressed in 90% (27 of 30) of patients with non–IG-ICC. The appearance of CDX2-positive epithelial cells in IPNL coincided with that of MUC2-positive epithelial cells using the double staining technique (Fig. 5). MUC5AC was strongly expressed in all patients with IPNL and was expressed in 9 of 12 well-differentiated and 3 of 9 moderately differentiated non–IG-ICCs. In contrast, none of 9 poorly differentiated non–IG-ICCs exhibited MUC5AC expression. EGFR was rarely expressed in IPNL; EGFR was predominantly expressed in poorly differentiated non–IG-ICCs (9 of 9). Reciprocal expression of MUC5AC and EGFR occurred in non–IG-ICCs, as described above (Table 5). Finally, over-expression of p53 was detected sporadically in IPNL (6%); p53 over-expression was detected in 37% of non–IG-ICCs.
Table 4. Immunohistochemical Study for Cholangiocarcinoma and Its Precursors
Non–IG-ICC (n = 30)
NOTE. Scoring system of immunohistochemical study, see Patients and Methods. Values are given as n (%).
Abbreviations: IPNL, intraductal papillary neoplasm of the liver; non–IG-ICC, non-intraductal growth type cholangiocarcinoma; TFF1, trefoil factor 1; EGFR, epidermal growth factor receptor.
Table 5. Expression of MUC5AC and EGFR in Non-Intraductal Growth-Type Chologiocarcinoma in Relation to Histological Differentiation
(n = 12)
(n = 9)
(n = 9)
Epidermal growth factor receptor
A large spectrum of IPNL has been presented in this study, ranging from benign disease to invasive malignancy, and including chronic inflammation-dysplasia-carcinoma in situ-invasive carcinoma. Hepatolithiasis was often associated with initiation of cholangiocarcinogenesis; the association of hepatolithiasis in IPNL types 1 and 2, IPNL types 3 and 4, and non–IG-ICC was 100%, 79%, and 64%, respectively. Some bias appeared to be present within this cohort. Diagnosing IPNL types 1 and 2 preoperatively was not possible; diagnosis of these lesions required demonstration of ultrastructures by microscopic examination of tissue. In this study, females more commonly had hepatolithiasis (68%), IPNL type 1 (67%), IPNL type 2 (76%), IPNL type 3 (77%), and IPNL type 4 (66%). A close relationship existed between hepatolithiasis and the development of IG-ICC. The time lag between the development of hepatolithiasis and IPNL was 6 to 8 years; we suggest that such a latent period is required for elicitation of an inflammation–cell repair–dysplasia–malignant transformation process.
Fever, chills, and jaundice were more pronounced in patients with IPNL than in those with non–IG-ICC. Intraductal neoplasia intermingled with tenacious mucus and hepatolithiasis are particularly liable to occlude both the extra-hepatic and segmental bile ducts, predisposing to suppurative cholangitis and jaundice. If possible, IG-ICC should be differentiated from its precursor lesions. The presence of anemia, mucobilia, and a patulous orifice of the sphincter of Oddi help to distinguish the former from the latter. Mucobilia is a striking feature; it makes the diagnosis of IPNL straightforward. Generalized dilatation of bile ducts, which is relatively greater in the hepatic segmental ducts, may facilitate making a surgical plan.13 Serum carcinoembryonic antigen, but not CA19-9, was of some value in differentiating IG-ICC from its precursor lesions.
Survival of patients with IG-ICC was superior to that of patients with non–IG-ICC in this series; this finding is in agreement with results of other investigations.1, 5, 14, 15 The mean survival of patients with IPNL types 3 and 4 and non–IG- ICC was 55.5 ± 17.1 months, 36.9 ± 6.3 months, and 15.8 ± 1.8 months, respectively. The survival of patients with IPNL types 3 and 4 was similar (log-rank test, P = .12). Several reasons may account for these findings. First, not all patients with IPNL type 3 and malignant and dysplastic tissue at the margins of surgical specimens underwent further resection of hepatic tissue, because of either lack of consent or the surgeon's preference. Second, multi-centric dysplastic biliary epithelium might be present in the remnant liver and may be difficult to detect using currently available diagnostic tools. Third, the number of patients with IPNL type 3 may have been too small to detect significant differences. The survival data suggest that intraductal carcinoma in situ might take a short time (perhaps 1-2 years) to evolve into an invasive lesion associated with rapid clinical deterioration.
The cdx2 homeobox gene plays an essential role during early development of the midgut.16, 17 Given that aberrant expression of CDX2 is associated with intestinal differentiation in Barrett's adenocarcinoma,18 gastric cancer,19 and certain rat cholangiocarcinomas,20 CDX2 may be involved in progression of IPNL toward intestinal differentiation. In this study, the incidence of CDX2 expression increased from that in hepatolithiasis alone to that in IPNL types 1 and 2; it was maximal in IPNL type 3 and less in IPNL type 4. It was not expressed in non–IG-ICC. Furthermore, the appearance of CDX2-positive neoplastic cells in IPNL reflected that of MUC2-positive neoplastic cells; thus, CDX2 expression may be preferentially localized in mucin-producing neoplasia. Recently, promoter activity of the 5′-flanking region of the MUC2 gene has been examined; this region, which comprises bases 171 to 228, contains an element bound by a transcriptional factor, CDX2.21 This indicates that CDX2 protein interacts with the muc2 promoter and activates the transcription of muc2.22 Increasing evidence suggests that trefoil factor family (TFF) peptides are important in mucosal defense and repair of the gastrointestinal tract.23 Of them, TFF1 is considered to be bi-functional, either a tumor suppressor24 or a mitogen.25 In the current study, the incidence of TFF1 expression increased with progression from hepatolithiasis alone, through IPNL types 1 and 2; it was maximal in IPNL type 3 and was less in IPNL type 4. In contrast to CDX2, expression of TFF1 was detected in differentiated non–IG-ICCs. Furthermore, MUC5AC expression, a marker of gastric foveolar metaplasia, was detected in all patients with IPNL, as well in 12 (40%) of patients with more differentiated non–IG-ICCs. We postulate that up-regulation of TFF1 and MUC5AC in hepatolithiasis, IPNL, and more differentiated non–IG-ICCs may reflect gastric metaplasia and neoplastic transformation. Effects of epidermal growth factor (EGF) are mediated through its specific cell-surface receptor (EGFR), which has certain structural similarities to the avian erythroblastosis virus v-erb-B transforming protein (ERBB).26–28 Over-expression of proto-oncogene-encoded receptor, tyrosine kinase ERBB-2, and up-regulation of cyclo-oxygenase-2 appear to be important in cholangiocarcinogenesis.29, 30 In the current study, EGFR expression was detected in 40% of patients with non–IG-ICC, but in only 3 (5%) of 59 patients with IPNL type 4 and in no patients with IPNL types 1 to 3. The reported frequencies of p53 over-expression in cholangiocarcinoma vary from 19% to 58%.31–34 In this study, p53 over-expression was detected less in IG-ICC than in non–IG-ICC (6% vs. 37%, P < .01). Expression of MUC2 and CDX2 is characteristic of IPNL, whereas over-expression of MUC1, EGFR, and p53 is rare in IPNL. TFF1 and MUC5AC are frequently expressed in both IPNL and well-differentiated non–IG-ICC.
In conclusion, a study of a cohort of patients with IG-ICC and its precursor lesions was conducted. Hepatolithiasis appeared to play an important role in initiating malignant transformation. Expression of CDX2, TFF1, and apomucin profile appears to modulate gastrointestinal metaplasia phenotypes of intraductal growth-type cholangiocarcinoma and its precursor lesions. Expression of CDX2 and MUC2 is helpful in differentiating IPNL and non–IG-ICC. Significant differences in survival are associated with the various stages of evolution of the lesions studied and justify our current aggressive surgical strategy.